Paper Title
Improving Bronchitis Diagnosis in Pediatrics: A Study of Deep Learning Models on Chest X-Ray Images

Abstract
Acute bronchitis is a prevalent respiratory condition in pediatrics, frequently resulting in hospitalizations and resource-intensive diagnostic procedures. The objective of this study is to develop a computer-aided diagnosis system that can automatically classify bronchitis in pediatric patients based on chest X-ray images, utilizing deep learning techniques. The dataset employed comprises pediatric chest X-ray images from the VinDr-CXR collection, encompassing 1016 cases of bronchitis and 1016 healthy controls. The performance of five deep learning architectures such as Xception, DenseNet-121, InceptionV4, EfficientNetV2B7, and ResNet-101, was evaluated in distinguishing bronchitis from control cases. Following rigorous training and testing procedures, including K-fold cross-validation, the classification performance of each model was evaluated based on accuracy, precision, F1 score, and Matthew’s correlation coefficient. The InceptionV4 model demonstrated the highest accuracy (0.943) and overall performance among the tested models, closely followed by EfficientNetV2B7. The findings demonstrate that deep learning models, particularly InceptionV4, have the potential to enhance the diagnostic process for bronchitis by reducing the workload of physicians and minimizing the possibility of diagnostic errors. This computer-aided diagnosis approach has the capacity to improve pediatric healthcare by facilitating rapid and precise bronchitis diagnosis. Future work will concentrate on integrating interpretability features and validating the efficacy of the model across diverse datasets. Keywords - Pediatric Bronchitis, Deep Learning, X-Ray Imaging, Computer-Aided Diagnosis